XRecommend Sandbox allows you to simulate the response of a DeepRec Model from a specific element and locale. The process revolves around creating an event sequence that mimics a customer’s journey. Choose from different types of events to create a string of events that inform the recommendation output. Unlike XSearch, all of the inferencing for XRecommend is happening in real-time, based on customer’s actions on your site.
Prerequisite
Experiences and XRecommend Engines must be created before they can be tested in the sandbox.
Current Events available for simulation include:
Product View
When a customer views a product page, it signals interest and provides valuable insights into product popularity, helping refine recommendations and optimize merchandising strategies across the entire catalog and customer base.
Add To Cart
Adding a product to the cart indicates a strong purchase intent and allows the system to trigger upsell or cross-sell recommendations, depending on your settings.
Product Purchase
A completed purchase confirms successful conversion and provides critical data for future product recommendations. Although a product may not be likely to be purchased again, in certain brand contexts, your model will take this into account and tailor product suggestions accordingly.
Search Products
To populate your unique customer journey, use the Search bar to input product codes or search terms. Once you select one or multiple products, items can be re-arranged to test outcomes.
Testing Your Recommendation Engine
Once you have composed your event sequence, click the “Test Engine” button in the right corner of your Event Composer.
Your recommendations will be populated above the composer tray in a feed format. Each response will have the locale, element name, time, and the event sequence listed. The recommendation carousel will appear below this information for you to click on and scroll through.
